-2
$\begingroup$

I have the number 1 00111 010 where there is one signed bit, 5 bits for exp and 3 for frac. I got that the value of it is -1/1024 but I'm not sure if that's right. I also need to get the format B of the representation where there is one signed bit, 4 exp bits, and 4 frac bits

  • 0
    You need to know the exponent bias in order to answer this accurately. I'm not aware that there actually are IEEE standards for 9-bit floating-point numbers. – 2017-02-09
  • 0
    You can just use the formula 2^(k-1)-1 to get the bias so for the original question the bias is 15 and for format B the bias is 7 – 2017-02-09
  • 0
    Is that because you said "IEEE"? The IEEE standard specifies not only a fixed bias for each floating-point format, for each number of bits in an encoding there is a fixed number of bits in the mantissa and a fixed number of bits in the exponent. You wouldn't find a 32-bit IEEE-754 number with a 9-bit exponent, for example. Perhaps someone is using the word "IEEE" to signify their own peculiar extension of the standard, but it would be better to tell us what that standard is than to assume we know it. – 2017-02-09
  • 0
    Just to give you an idea of how many different formats have used a sign-bit/exponent/mantissa format like IEEE, but with different numbers of bits, see http://www.quadibloc.com/comp/cp0201.htm -- and notice that in most of these formats the exponent bias is $2^{k-1}+1.$ – 2017-02-09

1 Answers 1

0

As the "IEEE" hints at the usual normalization, you have to add the leading $1$ to the mantissa. Your number is $$ (-1)^1·2^{7-15}·(1.010)_2 $$